All Detailer nodes, except for FaceDetailer and Detailer For AnimateDiff, do not allow image batch inputs. Because SEGS is already representing multiple parts within a single image, turning images into batches can lead to confusion. Therefore, we recommend handling this explicitly through List conversion.
- FaceDetailer internally handles SEGS only, so it has allowed batches since V4.50.
This tutorial explains how to convert Batch to List and List to Batch.
-
Batch processing can only be applied to the latent space and cannot be applied to the pixel image targeted by the detailer.
-
You can use the Rebatch Latents node to remove separate latent images from their batches.
-
Once the latent image is passed through the Rebatch Latents node, no batch processing can be done. Putting the node directly before VAE Decode will allow your primary samplers to run with batched latents and will only unbatch them before VAE decoding.
-
Use a batch_size of 1 to obtain unbatched latents.
- If you want to maintain as many batches as possible during the latent stage, you can convert the output, which is turned into an image through VAE Decode, into a format that FaceDetailer can handle using ImageBatchToImageList, just before the input to FaceDetailer.
- When you need to convert post-processed images, which have been converted into a list, back into a batch for handling, you can use the
Image List to Image Batchnode to convert them back into batch images and work with them using tools like the detailer.
